FUZZY TL-UNIFORM SPACES KHALED A. HASHEM AND NEHAD N. MORSI Received 12 September 2005; Revised 12 April 2006; Accepted 25 April 2006 The main purpose of this paper is to introduce a new structure that is a fuzzy TL-uniform space. We show that our structure generates a fuzzy topological space, precisely, a fuzzy Tlocality space. Also, we deduce the concept of level uniformities of a fuzzy TL-uniformity. We connect the category of fuzzy TL-uniform spaces with the category of uniform spaces. We establish a necessary and sufficient condition, under which a fuzzy TL-uniformity is probabilistic pseudometrizable. Finally, we define a functor from the category of fuzzy TL-uniform spaces into the category of fuzzy T-locality spaces and we show that it preserves optimal lifts. Copyright © 2006 Hindawi Publishing Corporation. All rights reserved. 1. Introduction In this paper, we introduce, for each continuous triangular norm T, a new structure that is a fuzzy TL-uniform space, which generates a fuzzy topological space. We proceed as follows. In Section 2, we present some basic definitions and ideas, also, we supply three lemmas on the α-cuts of fuzzy subsets on a universe set X for all α in the interval [0,1]. We need those lemmas for proofs scattered in subsequent sections. In Section 3, we introduce our definition of fuzzy TL-uniform spaces, fuzzy uniform map, and the fuzzy topology associated with a fuzzy TL-uniform space. In Section 4, we study the relationship between a fuzzy TL-uniformity and its α-level uniformities. Also, we generate fuzzy TL-uniformities from classical uniformities. In Section 5, we prove that each category of fuzzy TL-uniform spaces and uniform maps between them is a topological category, and we describe the fuzzy TL-uniform space of the optimal lift of a source in this category. Also, we define a functor from the category of fuzzy TL-uniform spaces into the category of fuzzy T-locality spaces, and we show that it preserves optimal lifts. Hindawi Publishing Corporation International Journal of Mathematics and Mathematical Sciences Volume 2006, Article ID 25094, Pages 1–24 DOI 10.1155/IJMMS/2006/25094 2 Fuzzy TL-uniform spaces 2. Prerequisites A triangular norm (cf. [1, 2]) is a binary operation on the unit interval I = [0,1] that is associative, symmetric, isotone in each argument and has neutral element 1. A continuous triangular norm T is uniformly continuous, because its domain I × I is compact. This means that for all ε > 0, there is θ = θT,ε > 0 in such a way that for every (α,β) ∈ I × I, we have (αTβ) − ε ≤ (α − θ)T(β − θ) ≤ αTβ ≤ (α + θ)T(β + θ) ≤ (αTβ) + ε. (2.1) Given a lower semicontinuous triangular norm T, the following binary operation on I: jT (α,γ) = sup{θ ∈ I : αTθ ≤ γ}, (α,γ) ∈ I × I, (2.2) is called the residual implication of T [3], we often simplify jT to j. The best wellknown triangular norms are the simplest three, namely Min (also denoted by ∧), Tm (the Lukasiewicz conjunction), and π (product), where for all α,β ∈ I, αTm β = (α + β)∧1, απβ = αβ, (2.3) where the binary operation ∧, above, is the truncated subtraction, defined on nonnegative real numbers by r ∧ s = max{r − s,0}, r,s ≥ 0. Proposition 2.1 [4]. The T-residual implication j, of a lower semicontinuous triangular norm T, enjoys the following properties for all α,β,γ ∈ I: (IT1) αTβ ≤ γ if and only if α ≤ j(β,γ), (IT2) αTβ > γ if and only if α > j(β,γ), (IT3) j(α,γ) = 1 if and only if α ≤ γ, (IT4) j(1,γ) = γ, (IT5) j(αTβ,γ) = j(α, j(β,γ)), (IT6) αT j(α,γ) ≤ γ, (IT7) j(α,β)T j(θ,γ) ≤ j(αTθ,βTγ), (IT8) j is antimonotone in the left argument and monotone in the right argument. A fuzzy set λ in a universe X, introduced by Zadeh in [5], is a function λ : X → I. The collection of all fuzzy subsets of X is denoted by I X . Ordinary subsets of a universe X will frequently occur in what follows. We will often need to consider a subset M ⊆ X as a fuzzy subset of X, said to be a crisp subset of X, which we will denote by the symbol 1M . We do this by identifying 1M with its characteristic function. The collection of all crisp subsets of X is denoted by 2X . The symbol α denotes the constant fuzzy set in a universe X with value α ∈ I. For the fuzzy sets μ,λ ∈ I X , the degree of containment of μ in λ according to j = jT is the real number in I [6]: j μ,λ = inf j μ(x),λ(x) . x ∈X (2.4) K. A. Hashem and N. N. Morsi 3 Last, we denote by j |μ,λ| the following fuzzy subset of X: j |μ,λ|(x) = j μ(x),λ(x) , x ∈ X. (2.5) Given a fuzzy set λ ∈ I X and a real number α ∈ I1 = [0,1[, the strong α-cut of λ is the following subset of X: λα = x ∈ X : λ(x) > α , (2.6) and for a real number α ∈ I, the weak α-cut of λ is the subset of X: λα∗ = x ∈ X : λ(x) ≥ α , (2.7) for both strong α-cuts and weak α-cuts of λ, we identify crisp sets with their characteristic functions. It is direct to verify that every λ ∈ I X has the following resolutions: λ= α ∧ λα = α∈I1 α ∧ λ α∗ . (2.8) α∈I0 We follow Lowen’s definition of a fuzzy interior operator on a set X [7]. This is an operator o : I X → I X that satisfies μo ≤ μ, (μ ∧ λ)o = μo ∧ λo for all μ,λ ∈ I X , and αo = α for all α ∈ I. The pair (X,o ) is called a fuzzy topological space (FTS). The category of all fuzzy topological spaces and continuous functions between them (cf. [7]) is denoted by FTS. Given a fuzzy topological space (X,o ) = (X,τ), and α ∈ I1 , the set ια (τ) = λα ⊆ X : λ ∈ τ (2.9) is a topology on X, said to be the α-level topology of τ [8]. It is direct to see that its interior operator is given by intια (τ) (M) = α ∨ 1M o α , M ⊆ X, (2.10) and, ι(τ) = supα∈I1 ια (τ) is called the topology modification of τ [8]. Prefilters and prefilterbases were introduced by Lowen in [9]. A prefilter in a universe / , is closed under finite meets X is a nonempty collection ⊂ I X which satisfies 0 ∈ and contains all the fuzzy supersets of its individual members. A prefilterbase in X is a / Ꮾ and the meets of two members of Ꮾ nonempty collection Ꮾ ⊂ I X which satisfies 0 ∈ contain a member of Ꮾ. A prefilterbase Ꮾ is said to be prefilterbasis for a prefilter if = [Ꮾ], where [Ꮾ] = {λ ∈ I X : there exists ν ∈ Ꮾ with ν ≤ λ}. 4 Fuzzy TL-uniform spaces Definition 2.2 [10]. The T-saturation operator is the operator ∼ T which sends a prefilterbase Ꮾ in X to the following subset of I X : ∼T Ꮾ ⎫⎤ ⎡⎧ ⎨ ⎬ ⎣ = γTμγ : μγ ∈ Ꮾ ∀γ ∈ [0,1[⎭⎦ , ⎩ (2.11) γ∈[0,1[ said to be the T-saturation of Ꮾ. Theorem 2.3 [10]. The T-saturation operator ∼ T is isotone. Also, given a prefilterbase Ꮾ in X, Ꮾ∼T satisfies Ꮾ∼T is a prefilter, ⊇ [Ꮾ] ⊇ Ꮾ, sup j υ,λ = sup j μ,λ, υ ∈Ꮾ μ ∈Ꮾ ∼ T (2.12) moreover, Ꮾ∼T equals the following subset of I X : λ ∈ I X : j β,λ ∈ [Ꮾ] ∀β < 1 . (2.13) Definition 2.4 [10]. A fuzzy T-locality space is a fuzzy topological space (X,o ) whose fuzzy interior operator is induced by some indexed family Ꮾ = (Ꮾ(x))x∈X , of prefilterbases in I X , in the following manner: μo (x) = sup j υ,μ, υ∈Ꮾ(x) ∀(μ,x) ∈ I X × X. (2.14) The family Ꮾ is said to be a T-locality basis for (X,o ), the fuzzy topology of (X,o ) will be denoted by τ(Ꮾ). The full subcategory, of FTS, of fuzzy T-locality spaces is denoted by T-FLS. Theorem 2.5 [10]. A family of prefilterbases in X, Ꮾ = (Ꮾ(x))x∈X , will be a T-locality base in X if and only if it satisfies the following two conditions for all x ∈ X. (TLB1) υ(x) = 1 for all υ ∈ Ꮾ(x). (TLB2) Every υ ∈ Ꮾ(x) has a T-kernel. This consists of two families (υγ ∈ Ꮾ(x))γ∈I1 and (υ yγθ ∈ Ꮾ(y))(y,γ,θ)∈X ×I1 ×I0 such that for all (y,γ,θ) ∈ X × I1 × I0 , γTυγ (y) ∧θ Tυ yγθ ≤ υ, where I0 =]0,1]. (2.15) Theorem 2.6 [10]. (i) A family Ꮾ = (Ꮾ(x))x∈X will be a T-locality basis for a T-locality space (X,o ) if and only if Ꮾ∼T is so. In this case, Ꮾ∼T is the greatest T-locality basis for (X,o ) (and Ꮾ∼T is called the T-locality system of (X,o )). It is the unique T-saturated T-locality basis for (X,o ). In this case, Ꮾ is a basis for the T-locality system Ꮾ∼T . (ii) If Ꮾ = (Ꮾ(x))x∈X is a T-locality basis for a T-locality space (X,o ), then Ꮾ∼T = ((Ꮾ(x))∼T )x∈X is obtained also by Ꮾ(x) ∼T = υ ∈ I X : υo (x) = 1 , x ∈ X. (2.16) A distance distribution function (DDF) [2] is a function from the set R+ of positive real numbers to the unit interval I = [0,1], which is monotone, left continuous, and has K. A. Hashem and N. N. Morsi 5 supremum 1. The set of all DDFs is denoted by D+ . Every continuous triangular norm T induces on D+ a binary operation ⊕T by (η ⊕T ζ)(s) = sup{η(b)Tζ(s − b) : 0 < b < s}, s > 0. The identity element of a semigroup (D+ , ⊕T ) is ε0 defined by ⎧ ⎨0, ε0 (x) = ⎩ 1, x ≤ 0, x > 0. (2.17) A probabilistic pseudometric on a set X (cf. [11]) is a function Γ : X × X → D+ that satisfies for all x, y, z in X the following properties: (PM1) Γ(x,x) = ε0 , (PM2) Γ(x, y) = Γ(y,x), (PM3) Γ(x, y) ⊕T Γ(y,z) ≥ Γ(x,z). In [12], Höhle defines for every ψ,ϕ ∈ I X ×X and λ ∈ I X , the T-section of ψ over λ by ψ λT (x) = sup y∈X λ(y)Tψ(y,x), x ∈ X, the T-composition of ψ, ϕ by ψoT ϕ(x, y) = supz∈X ϕ(x,z)Tψ(z, y), x, y ∈ X, the symmetric of ψ by s ψ(x, y) = ψ(y,x), x, y ∈ X. Now, we will use the following lemmas in the sequel. Lemma 2.7. For all ϕ, ψ ∈ I X ×X and α ∈ I1 , (i) (γTϕ)α = ϕ j(γ,α) for all γ ∈ I, (ii) (ϕoT ψ)α = θTβ>α (ϕθ ∗ oψβ∗ ) = θ β θTβ≤α (ϕ oψ ). Proof. (i) For every (x, y) ∈ X × X, we get the equivalences α (x, y) ∈ γTϕ iff γTϕ(x, y) > α iff ϕ(x, y) > j(γ,α), by (IT2), (2.18) iff (x, y) ∈ ϕ j(γ,α) . This proves (i). (ii) For the first equality, let (x, y) ∈ X × X, we get the equivalences (x, y) ∈ ϕoT ψ iff iff α ϕoT ψ (x, y) > α sup ψ(x,z)Tϕ(z, y) > α z ∈X iff ∃z ∈ X, β,θ ∈ Io with βTθ > α such that (2.19) ψ(x,z) ≥ β, ϕ(z, y) ≥ θ iff ∃z ∈ X, β,θ ∈ Io with βTθ > α such that (x,z) ∈ ψβ∗ , (z, y) ∈ ϕθ∗ iff (x, y) ∈ θTβ>α which proves our assertion. ϕθ∗ oψβ∗ . 6 Fuzzy TL-uniform spaces For the second equality of (ii), we get also (x, y) ∈ / ϕoT ψ iff iff α ϕoT ψ (x, y) ≤ α sup ψ(x,z)Tϕ(z, y) ≤ α z ∈X iff ∀z ∈ X, ∃β,θ ∈ I0 with βTθ ≤ α such that (2.20) ψ(x,z) ≤ β, ϕ(z, y) ≤ θ iff ∀z ∈ X, ∃β,θ ∈ I0 with βTθ ≤ α such that / ϕθ (x,z) ∈ / ψ β , (z, y) ∈ iff ϕθ oψ β . (x, y) ∈ / θTβ≤α This completes the proof. Lemma 2.8 [13]. For all ψ ∈ I X ×X and α ∈ I1 , sψ α = s ψα . (2.21) Lemma 2.9 [14]. Let (ψn )n∈N be a sequence of fuzzy sets on X × X for which ψn (x,x) = 1 for all x ∈ X; ψn = s (ψn ); ψn+1 oT ψn+1 oT ψn+1 ≤ ψn . Then, there is a probabilistic pseudometric Γ on X satisfying ψn+1 (x, y) ≤ Γ(x, y) 2−n ≤ ψn (x, y), ∀(x, y) ∈ X × X. (2.22) 3. Fuzzy TL-uniform spaces The fuzzy TL-uniform spaces are introduced in this section, and some of their properties are given. The uniform maps are defined. Also, we generate a fuzzy topology from a fuzzy TL-uniform space. Definition 3.1. (i) A fuzzy TL-uniform base on a set X is a subset υ ⊂ I X ×X which fulfills the following properties. (FLUB1) υ is a prefilterbase. (FLUB2) For all ϕ ∈ υ and x ∈ X, ϕ(x,x) = 1. (FLUB3) For all ϕ ∈ υ and γ ∈ I1 , there is ϕγ ∈ υ with γTϕγ ≤ s ϕ. (FLUB4) For all ϕ ∈ υ and γ ∈ I1 , there is ϕγ ∈ υ with γT(ϕγ oT ϕγ ) ≤ ϕ. Obviously, (FLUB3) and (FLUB4) can be replaced by the single condition (FLUB3\ ). For all ϕ ∈ υ and γ ∈ I1 , there is ϕγ ∈ υ with γT(ϕγ oT ϕγ ) ≤ s ϕ. (ii) A fuzzy TL-uniformity on X is a T-saturated fuzzy TL-uniform base on X. (iii) If μ is a fuzzy TL-uniformity on X, then υ is a basis for μ if υ is a prefilterbase and υ∼T = μ. K. A. Hashem and N. N. Morsi 7 It follows that for a fuzzy TL-uniformity μ on a set X, and all ϕ ∈ μ, s ϕ ∈ μ. The pair (X,μ) consisting of a set X and a fuzzy TL-uniformity μ on X is called fuzzy TL-uniform space. The elements of μ are called fuzzy vicinities. Proposition 3.2. Let υ be a fuzzy TL-uniform base on a set X, then the prefilter υ∼T determined by υ is a fuzzy TL-uniformity on X. Conversely, if μ is a fuzzy TL-uniformity on X, then conditions (FLUB2)–(FLUB4) are satisfied by every prefilterbase υ which determined μ. Proof. Suppose υ is a fuzzy TL-uniform base on X. We show that υ∼T is a fuzzy TLuniformity. Obviously, υ∼T is T-saturated prefilter. Now, to prove (FLUB2), let ϕ ∈ υ∼T , then there exists (ψα ∈ υ)α∈I1 such that ϕ ≥ α∈I1 (αTψα ). Hence for every x ∈ X, we have ϕ(x,x) ≥ αTψα (x,x) α∈I1 = (αT1) for ψα ∈ υ α∈I1 = (3.1) (α) = 1. α∈I1 Also, to prove (FLUB3\ ), let ϕ ∈ υ∼T and γ ∈ I1 , then for all α ∈ I1 , there exists ψα ∈ υ such that ϕ ≥ αTψα . (3.2) But for each β ∈ I1 , there is β ψα ∈ υ with β βT ψα oT β ψα ≤ s ψα . (3.3) By continuity of T, we can choose β ∈ I1 such that βTβ = γ, and by taking ϕγ = β ψβ ∈ υ ⊆ υ∼T , we get ≤ βT s ψβ , by (3.3), γT ϕγ oT ϕγ = βTβT β ψβ oT β ψβ ≤ s ϕ, (3.4) directly from (3.2). This proves that υ∼T is a fuzzy TL-uniformity. Conversely, suppose that μ is a fuzzy TL-uniformity, and let υ be a prefilterbase which determines μ. Since υ ⊆ μ, then obviously υ satisfies (FLUB2), while (FLUB3) follows immediately from the fact that for all ψ ∈ μ, we get s ψ ∈ μ. 8 Fuzzy TL-uniform spaces To prove (FLUB4), let ϕ ∈ υ and γ ∈ I1 , then ϕ ∈ μ and hence, there is a family (ψγ ∈ μ)γ∈I1 such that γT ψγ oT ψγ ≤ ϕ. (3.5) By continuity of T, we can find α ∈ I1 such that γ ≤ αTαTα, and by Theorem 2.6, there is ϕγ in υ with ϕγ ≤ j |α,ψα |, consequently, γT ϕγ oT ϕγ ≤ αTαTαT j α,ψα oT j α,ψα = αT ≤ αT ψα oT ψα , ≤ ϕ, αT j α,ψα oT αT j α,ψα , clear, (3.6) by (IT6), by (3.5), which proves that υ satisfies (FLUB4), and completes the proof. } ⊆ I X ×X Lemma 3.3. If {ψα : α ∈ I1 and ψ = X ψ 1M T = α∈I1 [αTψα 1M T ] ∈ I . α∈I1 (αTψα ) ∈ I X ×X , then for every M ∈ 2X , Proof. For every x ∈ X, ψ 1M T (x) = sup 1M (y)Tψ(y,x) = sup ψ(y,x) y ∈X y ∈M ⎫ ⎧ ⎨ ⎬ = sup αTψα (y,x) ⎭ ⎩ y ∈M = α∈I1 !" αT sup ψα (y,x) αTψα 1M = T α∈I1 Hence ψ 1M T = by isotonicity of T, , y ∈M α∈I1 (3.7) (x). α∈I1 [αTψα 1M T ]. Lemma 3.4. For a prefilterbase ψ 1M T # in X × X and M ∈ 2X , :ψ ∈ ∼T $ " = ψ 1M T :ψ ∈ # $ "∼T . Proof. First, we show that {ψ 1M T : ψ ∈ ∼T } is T-saturated prefilter. Let ψ,ϕ ∈ and let μ ∈ I X be such that μ ≥ ψ 1M T ∧ ϕ1M T . Then we define φ ∈ I X ×X by ⎧ ⎨μ(x) φ(x, y) = ⎩ 1 if x ∈ M, if x ∈ / M. (3.8) # ∼T (3.9) K. A. Hashem and N. N. Morsi 9 # It is easy to see that μ = φ1M T and φ ≥ ψ ∧ ϕ and hence φ ∈ ∼T . This proves that # # μ ∈ {ψ 1M T : ψ ∈ ∼T }, demonstrates that {ψ 1M T : ψ ∈ ∼T } is a prefilter. # ∼T # ∼T Now, given a subfamily {ψγ : γ ∈ I1 } of . Since is T-saturated, α∈I1 (αTψα ) ∈ #∼T , so, αTψα 1M α∈I1 T = αTψα 1M α∈I1 ∈ ψ 1M :ψ ∈ T T, ∼T $ by Lemma 3.3, " (3.10) . This proves our assertion. Finally, we have ψ 1M T :ψ ∈ $ ⊆ ψ 1M T :ψ ∈ ∼T $ " $ with γTψγ ≤ ψ = ψ 1M T : ∀γ ∈ I1 ∃ψγ ∈ $ with γTψγ 1M T ≤ ψ 1M T ⊆ ψ 1M T : ∀γ ∈ I1 ∃ψγ ∈ $ ∼T , = ψ 1M T : ψ ∈ (3.11) since the prefilter {ψ 1M T : ψ ∈ ψ 1M T # ∼T :ψ ∈ } is T-saturated, then $ ∼T ⊆ ψ 1M ⊆ ψ 1M T T :ψ ∈ :ψ ∈ ∼T $ " $ ∼T (3.12) . So, equality holds, and winds up the proof. Theorem 3.5. If υ is a fuzzy TL-uniform base on a set X, then the indexed family Ꮾ = (Ꮾ(x))x∈X , given by Ꮾ(x) = ϕ 1x T :ϕ∈υ , is a T-locality base on X. (3.13) Proof. Given x ∈ X and υ ∈ Ꮾ(x), then there is ϕ ∈ υ such that υ = ϕ1x T , hence υ(x) = ϕ 1x T (x) = sup y ∈X It follows that Ꮾ satisfies (TLB1). 1x (y) Tϕ(y,x) = 1x (x)Tϕ(x,x) = 1. (3.14) 10 Fuzzy TL-uniform spaces Also, for every υ ∈ Ꮾ(x), γ ∈ I1 , and z ∈ X, we have υ(z) = ϕ 1x T (z) for some ϕ ∈ υ = ϕ(x,z) ≥ γT ϕγ oT ϕγ (x,z), = sup γTϕγ (x, y)Tϕγ (y,z) by (FLUB4) y ∈X ≥ sup γTϕγ (x, y)T δT ϕγδ oT ϕγδ (y,z) , y ∈X = sup γTϕγ (x, y)TδT sup ϕγδ (y,s)Tϕγδ (s,z) y ∈X by (FLUB4) again (3.15) s∈X ≥ γTϕγ (x, y)TδTϕγδ (y,z)Tϕγδ (z,z) = γTϕγ 1x T (y)Tδ T ϕγδ 1 y T (z) . Now, for every θ ∈ I0 , we can get (by continuity of T) δ = δθ ∈ I1 for which [γTϕγ 1x T (y)Tδ] ≥ [(γTϕγ 1x T (y))∧θ], hence take υxγ = ϕγ 1x T ∈ Ꮾ(x) and υ yγθ = ϕγδ 1 y T ∈ Ꮾ(y) which satisfy υ(z) ≥ [(γTυxγ (y))∧θ]Tυ yγθ (z). This shows that Ꮾ satisfies (TLB2), which completes the proof. Proposition 3.6. If μ is a fuzzy TL-uniformity on a set X, then the indexed family Ꮾ = (Ꮾ(x))x∈X , given by Ꮾ(x) = ϕ 1x T :ϕ∈μ , is a T-locality system on X. (3.16) The proof follows at once from Theorem 3.5 and Lemma 3.4. We will see how a fuzzy TL-uniform base can generate a fuzzy topology. Definition 3.7. The fuzzy T-locality space (X,τ T (υ)) induced by a fuzzy TL-uniform base υ on X is the T-FLS defined by means of the fuzzy T-locality base of the preceding theorem, for which its fuzzy interior operator o is given by % λo (x) = sup j ϕ 1x ϕ ∈υ T ,λ & , (λ,x) ∈ I X × X. (3.17) The fuzzy topology τ T (υ) of that space is also denoted simply by τ(υ). Notice that due to Theorem 1.4, a fuzzy TL-uniform base υ and its fuzzy TL-uniformity υ∼T induce the same fuzzy T-locality space, that is, τ(υ) = τ(υ∼T ). Uniform-type continuity: Definition 3.8. Let (X,μ) and (X \ ,μ\ ) be fuzzy TL-uniform spaces and f : X → X \ . f is a fuzzy uniform map (or fuzzy uniformly continuous) if any of the following equivalent conditions hold. (i) For each ψ ∈ μ\ , ( f × f )−1 (ψ ) ∈ μ. (ii) For each ψ ∈ μ\ , there is ψ ∈ μ such that ψ ≤ ( f × f )−1 (ψ ). (iii) For each ψ ∈ μ\ , there is ψ ∈ μ such that ( f × f )(ψ) ≤ ψ , where ( f × f )−1 (ψ )(x, y) = ψ ( f (x), f (y)), x, y ∈ X. K. A. Hashem and N. N. Morsi 11 The composite of two uniformly continuous functions f : (X,μ) → (Y ,U) and g : (Y ,U) → (Z, ℘) is again uniformly continuous, since (g f × g f )−1 (ψ) = ( f × f )−1 {(g × g)−1 (ψ)} ∈ μ, for all ψ ∈ ℘. Proposition 3.9. If (X,μ) and (X \ ,μ\ ) are fuzzy TL-uniform spaces, with bases υ and υ\ , respectively, and f : X → X \ , then f is uniformly continuous if and only if for all ϕ ∈ υ\ and all γ ∈ I1 , there is ϕ ∈ υ such that γTϕ ≤ ( f × f )−1 (ϕ ). (3.18) Proof. Let f : (X,μ) → (X \ ,μ\ ) be uniformly continuous and ϕ ∈ υ\ , γ ∈ I1 . Then ϕ ∈ υ\∼T = μ\ , since f is uniformly continuous, we get ( f × f )−1 (ϕ ) ∈ μ = υ∼T , hence there is ϕ ∈ υ such that γTϕ ≤ ( f × f )−1 (ϕ ). Conversely, suppose that the stated condition holds. Let ψ ∈ μ\ , since μ\ = υ\∼T , then for all γ ∈ I1 , there exists ϕγ ∈ υ\ such that ψ ≥ γTϕγ , hence ( f × f )−1 (ψ ) ≥ ( f × f )−1 γTϕγ = γT ( f × f )−1 ϕγ ≥ γTθTϕγθ for some θ ∈ I1 , ϕγθ ∈ υ, by hypothesis. (3.19) Since T is continuous, then for all α ∈ I1 , there exists γ = γα ∈ I1 such that α = γTγ, by taking φα = ϕγγ ∈ υ, we get ( f × f )−1 (ψ) ≥ γTγTϕγγ = αTφα , (3.20) which implies that ( f × f )−1 (ψ) ∈ μ. Hence f is uniformly continuous, and winds up the proof. A function f : (X, o ) = (X,τ) → (Y , o ) = (Y ,τ \ ), between two fuzzy topological spaces, is said to be continuous [7]; if f −1 (μ) ∈ τ for all μ ∈ τ \ , μ ∈ I Y . Equivalently, if o f −1 (λo ) ≤ [ f −1 (λ)] for all λ ∈ I Y . Theorem 3.10. Let (X,μ) and (Y , ℘) be fuzzy TL-uniform spaces. If f : X → Y is uniformly continuous, then it is continuous with respect to the fuzzy topologies generated by μ and ℘, respectively. Proof. Let λ ∈ I Y and x ∈ X, we denote by o1 and o2 the fuzzy interior operators of τ(μ) and τ(℘), respectively. Then, we have % f −1 λo2 (x) = λo2 f (x) = sup j ψ 1 f (x) ψ ∈℘ = sup inf j ψ 1 f (x) ψ ∈℘ y ∈Y T ,λ & T (y),λ(y) ≤ sup inf j ψ f (x), f (z) ,λ f (z) ψ ∈℘ z∈X for range f ⊆ Y 12 Fuzzy TL-uniform spaces = sup inf j ( f × f )−1 (ψ)(x,z), f −1 (λ)(z) ψ ∈℘ z∈X ≤ sup inf j ϕ(x,z), f −1 (λ)(z) , ϕ ∈μ z ∈X by uniformly continuous of f and (IT8), = sup inf j ϕ 1x T (z), f −1 (λ) (z) ϕ ∈μ z ∈X o1 = f −1 (λ) (x), (3.21) which proves the continuity of f : (X,τ(μ)) → (Y ,τ(℘)) at each point x in X. 4. The operators ιu,α and ωu Now, we introduce the concept of α-level uniformities for a given fuzzy TL-uniformity and we study the relationships between them. For a fuzzy TL-uniformity μ on a set X and α ∈ I1 , we define ιu,α (μ) = ψ j(β,α) ⊆ X × X : ψ ∈ μ, β ∈]α,1] , (4.1) we next show that ιu,α (μ) is a uniformity on X whenever μ is a fuzzy TL-uniformity, called the α-level uniformity of μ. In the following two propositions, let υ be a basis for a fuzzy TL-uniformity μ on a nonempty set X. Proposition 4.1. ιu,α (υ) is a filterbasis for the filter ιu,α (μ). Proof. We first show that ιu,α (υ) is indeed a filterbase. / ιu,α (υ), because every member in ιu,α (υ) contains the diagonal of X(D(X)). (a) ∅ ∈ (b) The intersection of two members of ιu,α (υ) contains a member: given U,V ∈ ιu,α (υ), then there are ψ,ϕ ∈ υ and β1 ,β2 ∈]α,1] such that U = ϕ j(β1 ,α) , V = ψ j(β2 ,α) , without loss of generality, β1 ≤ β2 . So U ∩ V = ϕ j(β1 ,α) ∩ ψ j(β2 ,α) ⊇ ϕ j(β1 ,α) ∩ ψ j(β1 ,α) , = (ϕ ∧ ψ) j(β1 ,α) because j β2 ,α ≤ j β1 ,α (4.2) . But, there is ζ ∈ υ with ζ ≤ ϕ ∧ ψ (because υ is a prefilterbase). Thus U ∩ V contains a member ζ j(β1 ,α) of ιu,α (υ). To prove that ιu,α (μ) is a filter, it now suffices to show that ιu,α (μ) is closed under supersets: given U ∈ ιu,α (μ), let V ⊇ U. Then U = ψ j(β,α) for some ψ ∈ μ, β ∈]α,1]. (4.3) K. A. Hashem and N. N. Morsi 13 If we take ϕ = V and j(β,α) ∈ I X ×X , we get ϕ ≥ ψ j(β,α) V = V ∨ j(β,α) j(β,α) j(β,α) ≥ ψ therefore, ϕ is also in μ = ϕ j(β,α) ∈ ιu,α (μ), (4.4) which proves our assertion. Finally, we show that ιu,α (μ) is generated by ιu,α (υ): let U ∈ ιu,α (μ), then U = ψ j(β,α) for some ψ ∈ μ and β ∈]α,1]. Since υ ⊆ υ∼T ⊂ μ, then for all γ ∈ I1 , there exists ϕγ ∈ υ with γTϕγ ≤ ψ. (4.5) By continuity of T, we can choose γo ∈ I1 such that γo Tβ ∈]α,1]. Then ϕγo j(γo Tβ,α) j(γo , j(β,α)) = ϕγo , j(β,α) = γo Tϕγo , ⊆ ψ j(β,α) , by (IT5), by Lemma 2.7(i), (4.6) by (4.5), = U. Thus there is a member (ϕγo ) j(γo Tβ,α) of ιu,α (υ) contained in U. Since also ιu,α (υ) is included in ιu,α (μ). This completes the proof that it is a filterbasis for ιu,α (μ). Proposition 4.2. ιu,α (υ) is a uniform basis for the uniformity ιu,α (μ). Proof. ιu,α (υ) is a uniform base. (UB1) ιu,α (υ) is a filterbase (from Proposition 4.1). (UB2) Each member of ιu,α (υ) evidently contains the diagonal D(X): since ϕ(x,x) = 1 > j(β,α) for all (ϕ,x,β) ∈ υ × X ×]α,1], then (x,x) ∈ ϕ j(β,α) for all (ϕ j(β,α) ,x) ∈ (ιu,α (υ)) × X. (UB3) For every U ∈ ιu,α (υ), its symmetric s U contains a member of ιu,α (υ): given U ∈ ιu,α (υ), then there exist ϕ ∈ υ and β ∈]α,1], such that U = ϕ j(β,α) , hence for all γ ∈ I1 , there exists ϕγ ∈ υ with γTϕγ ≤ s ϕ. (4.7) By continuity of T, we can get γo ∈ I1 such that γo Tβ ∈]α,1]. Then ϕγo j(γo Tβ,α) j(γo , j(β,α)) = ϕγo , j(β,α) = γo Tϕγo , j(β,α) ⊆ (s ϕ) , j(β,α) =s ϕ , by (IT5), by Lemma 2.7(i), by (4.7), by Lemma 2.8, = s U. Therefore, s U contains the member (ϕγo ) j(γo Tβ,α) of ιu,α (υ). (4.8) 14 Fuzzy TL-uniform spaces (UB4) For every U ∈ ιu,α (υ), there exists V ∈ ιu,α (υ) such that V oV ⊂ U: given U ∈ ιu,α (υ), there are ϕ ∈ υ and β ∈]α,1], such that U = ϕ j(β,α) . Also, for all γ ∈ I1 , there exists ϕγ ∈ υ with γT ϕγ oT ϕγ ≤ ϕ. (4.9) By continuity of T, we can choose γo ∈ I1 such that γo ≥ β and γo Tβ ∈]α,1]. Then U = ϕ j(β,α) ⊇ γ T ϕγo oT ϕγo j(β,α) o j(γo , j(β,α)) = ϕγo oT ϕγo , by Lemma 2.7(i), j(γo Tβ,α) = ϕγo oT ϕγo , by (IT5), θ δ = ϕγo o ϕγo , by Lemma 2.7(ii), θTδ ≤ j(γo Tβ,α) ⊇ θ ϕγo o ϕγo δ , (4.10) by (IT8), θTδ ≤ j(γo Tβ,αTα) j(γo ,α) j(β,α) ⊇ ϕγo o ϕγo , by (IT7), j(β,α) j(β,α) ⊇ ϕγo o ϕγo , by (IT8) for γo ≥ β. Taking V = (ϕγo ) j(β,α) , which is in ιu,α (υ), proves our assertion. This completes the proof that ιu,α (υ) is a uniform base. Consequently, by Proposition 4.1, ιu,α (μ) is uniformity with ιu,α (υ) a basis. Corollary 4.3. Let (X,μ) be a fuzzy TL-uniform space, in the cases T = Tm , min, the level uniformities of (X,μ) form an ascending chain, while in the case T = π, the level uniformities form an antichain. Proof. Let 0 ≤ α1 < α2 < 1 and V ∈ ιu,α1 (μ). Then there are ϕ ∈ μ and β ∈]α1 ,1] such that V = ϕ j(β,α1 ) . First, whenever T = Tm , we have j β,α1 = min 1 − β + α1 ,1 , by (2.2), = min 1 − β + α1 − α2 + α2 ,1 = j α2 + β − α1 ,α2 , (4.11) hence, V = ϕ j(β,α1 ) = ϕ j(α2 +β−α1 ,α2 ) ∈ ιu,α2 (μ), because it is easy to see that α2 + β − α1 ∈ ]α2 ,1], since β > α1 , which proves that ιu,α1 (μ) ⊆ ιu,α2 (μ). Second, whenever T = Min, we have j(β,α) = α ∀β > α, by (2.2). (4.12) K. A. Hashem and N. N. Morsi 15 Hence V = ϕ j(β,α1 ) = ϕα1 ⊇ ϕα2 , =ϕ because α1 < α2 , j(γ,α2 ) (4.13) for any γ ∈ α2 ,1 . But for every γ ∈]α2 ,1], ϕ j(γ,α2 ) ∈ ιu,α2 (μ), V ∈ ιu,α2 (μ), because ιu,α2 (μ) is a filter. Hence ιu,α1 (μ) ⊆ ιu,α2 (μ). Finally, whenever T = π, we have j(β,α) = α β ∀β > α, by (2.2). (4.14) Let W ∈ ιu,α2 (μ). Then W = ψ j(βo ,α2 ) for some ψ ∈ μ and βo ∈]α2 ,1]. For this number βo , take γ0 = α1 (βo /α2 ), we get W = ψ j(βo ,α2 ) = ψ (α2 /βo ) = ψ (α1 /γ0 ) = ψ j(γo ,α1 ) ∈ ιu,α1 (μ) (4.15) because, obviously, we can see that γo ∈]α1 ,1], which proves that ιu,α2 (μ) ⊆ ιu,α1 (μ). Proposition 4.4. If (X,μ) and (Y , ℘) are fuzzy TL-uniform spaces, and f : X → Y is a uniform map, then it is also a uniform map when considered as a function between the uniform spaces (X,ιu,α (μ)) and (Y ,ιu,α (℘)) for all α ∈ I1 . The proof immediately follows from the definition of α-level uniformities and the fact that ( f × f )−1 (ψ β ) = (( f × f )−1 (ψ))β for all ψ ∈ ℘, β ∈ I1 . For every fixed α ∈ I1 , define a function ι∼ u,α from the category of fuzzy TL-uniform spaces and fuzzy uniform maps to the category of uniform spaces and uniform maps by on objects : ι∼ u,α on morphisms : ι∼ u,α is ιu,α , is the identity function. (4.16) Then an obvious conclusion from the above proposition is that these ι∼ u,α are well-defined functors. Now, for a uniformity u on a set X, we define ωu (u) = ψ ∈ I X ×X : ∀γ ∈ I1 , ψ γ ∈ u , (4.17) we supply the proof that ωu (u) is a fuzzy TL-uniformity on X. Proposition 4.5. If u is a uniformity on a set X, then ωu (u) is a fuzzy TL-uniformity with u as a basis. Proof. First, it is easy to see that ωu (u) is a fuzzy TL-uniform base. To show that ωu (u) is T-saturated, let ψ ∈ (ωu (u))∼T . Then ∀γ ∈ I1 ∃ϕγ ∈ ωu (u) with ψ ≥ γTϕγ . (4.18) 16 Fuzzy TL-uniform spaces For every α ∈ I1 , choose γ = γα > α in I1 for which ψ α ⊇ γTϕγ α j(γ,α) = ϕγ , (4.19) since (ϕγ ) j(γ,α) ∈ u, because j(γ,α) ∈ I1 , then ψ α ∈ u because u is a filter, hence ψ ∈ ωu (u). This shows that (ωu (u))∼T = ωu (u). Second, to show that u is a basis for ωu (u), we prove that (u)∼T = ωu (u): let ψ ∈ (u)∼T . Then there is a family (Wβ ∈ u)β∈I1 such that ψ ≥ βT1Wβ . (4.20) Now, for every α ∈ I1 , we choose γ > α in I1 for which we get ψ α ⊇ γT1Wγ α α = γ ∧ 1W γ for 1Wγ is crisp (4.21) = 1Wγ = Wγ ∈ u, thus ψ α ∈ u, because u is filter, consequently, ψ ∈ ωu (u) renders (u)∼T ⊆ ωu (u). On the other hand, if ϕ ∈ ωu (u), then ϕ= α ∧ ϕα ≥ α∈I1 αTϕα . (4.22) α∈I1 Since ϕα ∈ u, for all α ∈ I1 , we get ϕ ∈ (u)∼T , therefore, ωu (u) ⊆ (u)∼T , and hence equal ity holds. Proposition 4.6. If (X,u) and (Y ,ᐃ) are uniform spaces, and f : X → Y is uniformly continuous in the usual sense, then it is uniformly continuous when considered as a function between the fuzzy TL-uniform spaces (X,ωu (u)) and (Y ,ωu (ᐃ)). Now, if we denote by TL-FUS (US) the category of fuzzy TL-uniform spaces (the category of uniform spaces) together with the uniformly continuous functions between these spaces, and define the map ω∼ : US → TL-FUS by setting ω∼ (X,u) = (X,ωu (u)) and ω∼ ( f ) = f , we get that ω∼ is a well-defined functor. In the following, we denote the fuzzy topology associated with a fuzzy TL-uniformity μ by τ(μ) and the topology associated with a uniformity u by T(u). Now, we introduce some compatibility between the above notions. Theorem 4.7. (i) The α-level topology ια (τ(μ)) of the fuzzy T-locality space (X,τ(μ)), coincides with the topology T(ιu,α (μ)) on X, induced by the α-level uniformity ιu,α (μ) of a fuzzy TL-uniform space (X,μ). (ii) For a uniform space (X,u), the fuzzy topology τ(ωu (u)) induced by the above fuzzy TL-uniformity ωu (u) is the same fuzzy topology ω(T(u)) ∼ (iii) ι∼ u,α oω = IdUS for all α ∈ I1 . K. A. Hashem and N. N. Morsi 17 Proof. (i) Let o be the fuzzy interior operator of the T-FLS (X,τ(μ)). Then for every nonempty M ∈ 2X , intια(τ(μ)) M = α∨M o α , by (2.10), o = x : α ∨ M (x) > α ( ' = x : sup inf j ψ 1x T (y), α ∨ M (y) > α , ψ ∈μ y ∈X by Definition 2.4, = x : ∃ψ ∈ μ, β ∈]α,1[ s.t. ∀ y ∈ / M, j ψ(x, y),α ≥ β / M, βTψ(x, y) ≤ α , by (IT1), = x : ∃ψ ∈ μ, β ∈]α,1[ s.t. ∀ y ∈ α = x : ∃ψ ∈ μ, β ∈]α,1[ s.t. ∀ y ∈ / M, (x, y) ∈ / βTψ (4.23) / M, (x, y) ∈ / ψ j(β,α) = x : ∃ψ ∈ μ, β ∈]α,1[ s.t. ∀ y ∈ / M, (x, y) ∈ /V = x : ∃V ∈ ιu,α (μ) s.t. ∀ y ∈ / M, y ∈ / V x = x : ∃V ∈ ιu,α (μ) s.t. ∀ y ∈ = x : V x ⊆ M for some V ∈ ιu,α (μ) = intT(ιu,α (μ)) M. This demonstrates that ια (τ(μ))) = T(ιu,α (μ)), which renders (i). (ii) Let the fuzzy set λ be τ(ω(u))-open. If x ∈ / (intT(u) λε ) for some x ∈ X and ε ∈ I1 , we get V x ⊂ λε , for all V ∈ u, thus, there / λε , that is, there exists z ∈ X such exists z ∈ X such that z ∈ V x for all V ∈ u with z ∈ that z ∈ V x for all V ∈ u with λ(z) ≤ ε, hence λ(x) = intτ(ω(u)) λ (x), by hypothesis = sup inf j ψ 1x ψ ∈ω(u) y ∈X T (y),λ(y) = sup inf j 1V 1x T (y),λ(y) , V ∈u y ∈X ≤ sup j 1V 1x T (z),λ(z) by Proposition 4.4, (4.24) V ∈u = j 1,λ(z) = λ(z), by (IT4), ≤ ε. That is, x ∈ / λε , which shows that λε ⊆ intT(u) λε , ε ∈ I1 . Consequently, λε = intT(u) λε , that ε is, λ ∈ T(u) for all ε ∈ I1 . 18 Fuzzy TL-uniform spaces On the other hand, let λ be ω(T(u))-open. Given x ∈ X, then for all ε ∈ I1 with ε < λ(x), we have x ∈ λε , and hence, by hypothesis x ∈ intT(u) λε , thus there exists W ∈ u such that W x ⊂ λε , that is, there exists W ∈ u such that for all z ∈ W x, we get z ∈ λε , that is, there exists W ∈ u such that for all z ∈ W x, we get λ(z) > ε. Hence, intτ(ω(u)) λ (x) = sup inf j ψ 1x ψ ∈ω(u) y ∈X T (y),λ(y) = sup inf j 1V 1x T (y),λ(y) (4.25) V ∈u y ∈X ≥ inf j 1W 1x T (y),λ(y) ≥ ε. y ∈X This shows that λ ≤ intτ(ω(u)) λ, which proves that λ is τ(ωu (u))-open, and winds up the proof of (ii). (iii) Follows immediately from the definitions. Example 4.8. Let (X,<) be an ordered set. For each point xo ∈ X, let Vxo = {(x, y) : x, y > xo } and define Uxo = D(X) ∪ Vxo . Then the reader can easily check that the collection u∗ = {Ux : x ∈ X } is a basis for a uniformity u on X. Hence by the above notions, we have ωu (u) = ψ ∈ I X ×X : ∀γ ∈ I1 , ψ γ ∈ u (4.26) is a fuzzy TL-uniformity on X with u as a basis. Moreover, ιu,α ωu (u) = ψ j(β,α) ⊆ X × X : ω ∈ ωu (u), β ∈]α,1] = u. (4.27) Definition 4.9. A fuzzy topological space (X,τ) is called a fuzzy TL-uniformizable if there is a fuzzy TL-uniformity μ on X such that τ = τ(μ). Corollary 4.10. A topological space (X,T) is uniformizable if and only if (X,ω(T)) is fuzzy TL-uniformizable. Proof. If T = T(u), then ω(T) = ω T(u) = τ ω(u) . (4.28) Conversely, let ω(T) = τ(μ), then T = ι ω(T) = sup ια ω(T) , α∈I1 = sup ια τ(μ) , α∈I1 = sup T ιu,α (μ) , α∈I1 ) * = T sup ιu,α (μ) . α∈I1 by definition of modification topology, by hypothesis, by Theorem 4.7(i), (4.29) K. A. Hashem and N. N. Morsi 19 Definition 4.11. A T-fls (X,τ(Ꮾ)) is said to be the following. (i) L-T0 , if for every x = y in X, there is υ ∈ Ꮾ(x) ∪ Ꮾ(y) such that υ(x) ∧ υ(y) < 1. (4.30) (ii) L-regular, if for every (M,x,ε) ∈ 2X × X × I0 , there is υ ∈ Ꮾ(x) with sup y∈X (υ ∧ 1M )(y) < ε, then there are an open set μ and ρ ∈ Ꮾ(x) such that 1M ≤ μ and sup y∈X (μTρ)(y) < ε. Theorem 4.12. TL-uniformizability ⇒ L-regularity. Proof. Suppose (X,μ) is a fuzzy TL-uniform space. Let M ∈ 2X , x ∈ X, and ε ∈ I0 are such that there is υ ∈ Ꮾ(x) with sup υ ∧ 1M (y) < ε, (4.31) y ∈X consequently, we can find ε0 very small such that sup υ ∧ 1M (y) + ε0 < ε. y ∈X (4.32) Since, υ ∈ Ꮾ(x), then for all γ ∈ I1 there are ψ,ψγ ∈ μ such that υ = ψ 1x T and ψ ≥ γT(ψγ oT ψγ ), (by using (FLUB4)). Hence ε > sup ψ 1x z ∈X ∧ 1 (z) + ε0 M T ≥ sup γT ωγ oT ψγ 1x T (z) + ε0 z ∈M = sup γT ψγ oT ψγ (x, y) + ε0 z ∈M ' = γT sup sup ψγ (x, y)Tψγ (y,z) ( z ∈M y ∈X + ε0 (4.33) ' ( = γT sup sup ψγ (x, y)T s ψγ (z, y) + ε0 y ∈X z ∈M ⎡ ≥ (γ + θ)T sup ⎣ψγ 1x y ∈X T (y)T z ∈M s ψγ 1z T (y) ⎤ ⎦, θ = θT,ε0 > 0 as in (2.1). 20 Fuzzy TL-uniform spaces Choosing γ0 ∈ I1 for which (γ0 + θ) = 1, and taking ρ = s ψγ0 1x T , μ = ( z∈M s ψγ0 1z T )o , we get ρ ∈ Ꮾ(x), and μ is open, with ε > sup y∈X [ρ(y)Tμ(y)], also μ ≥ 1M , because for every x ∈ M, we have ⎛ μ(x) = ⎝ ≥ s ψγ0 1z z ∈M s ψγ0 o o 1x = υx (x) = 1, ⎞o T T ⎠ (x) (4.34) (x) for s ψγ0 ∈ μ by Theorem 2.3. This proves the L-regularity of (X,τ(μ)). Theorem 4.13. If (X,μ) is a fuzzy TL-uniform space, then (X,τ(μ)) is L-T0 if and only if (inf ψ ∈μ ψ)1∗ = D(X). Proof. A T-FLS (X,τ(μ)) is L-T0 iff ∀x = y in X, ∃υ ∈ Ꮾ(x) ∪ Ꮾ(y) such that υ(x) ∧ υ(y) < 1 iff iff ∀x = y in X, ∃ψ ∈ μ such that ψ 1x T (x) ∧ ψ 1x T (y) < 1 ∀x = y in X, ∃ψ ∈ μ such that ψ(x, y) = ψ 1x T (y) < 1 iff ∀x = y in X, ∃ψ ∈ μ such that (x, y) ∈ / ψ1∗ iff inf ψ ∈μ 1∗ = / ψ ∈μ (4.35) ψ1∗ = D(X). Theorem 4.14. Let Γ be a probabilistic pseudometric on a set X. Then there is a fuzzy TLuniformity μ = μ(Γ) given by μ(Γ) = ψ ∈ I X ×X : ∀γ ∈ I1 ∃n ∈ N with γTΓ(x, y) 2−n ≤ ψ(x, y) . (4.36) Proof. Obviously, μ(Γ) is T-saturated prefilter. Now, if we put ψn (x, y) = Γ(x, y)(2−n ), we get a sequence (ψn )n∈N satisfies ψn (x,x) = 1, ψn = s (ψn ), and ψn+1 oT ψn+1 ≤ ψn , which shows that μ(Γ) is a fuzzy TL-uniformity. Remark 4.15. Let μ be a fuzzy TL-uniformity on X, a nonempty subset Ꮾ of μ is said to be a base of μ if for each ψ ∈ μ and each n ∈ N, there is ϕn ∈ Ꮾ such that (1 − 1/n)Tϕn ≤ ψ. Theorem 4.16. A fuzzy TL-uniformity μ on a set X is probabilistic pseudo-metrizable if and only if it has a countable base. Proof. Let μ be probabilistic pseudo-metrizable. Then by using (PM1) and Theorem 4.14, we get the necessary condition. For the sufficiency of the condition, suppose that μ has a countable base Ꮾ = {ψ1 ,ψ2 ,...,ψn ,...}. We set ϕ1 = ψ1 ∧ s ψ1 , then ϕ1 is a symmetric member of μ. K. A. Hashem and N. N. Morsi 21 For a fixed γ in I1 , apply (FLUB4) twice for the member ϕ1 ∧ ψ2 which is in μ, then there is a symmetric member say ϕ2 in μ such that γT ϕ2 oT ϕ2 oT ϕ2 ≤ ϕ1 ∧ ψ2 . (4.37) Next, consider ϕ2 ∧ ψ3 and get a symmetric ϕ3 in μ such that γT ϕ3 oT ϕ3 oT ϕ3 ≤ ϕ2 ∧ ψ3 . (4.38) In this manner, we proceed by induction a sequence (ϕn )n∈N of fuzzy vicinities for which ϕn = s ϕn , γT ϕn+1 oT ϕn+1 oT ϕn+1 ≤ ϕn ∧ ψn+1 ≤ ϕn , (4.39) by using Lemma 2.9, there is a probabilistic pseudometric Γ on X, with γTϕn+1 (x, y) ≤ Γ(x, y) 2−n ≤ ϕn (x, y); that is, the fuzzy TL-uniformity μ is generated by Γ, which proves our assertion. (4.40) 5. Optimal lifts in fuzzy TL-uniform spaces We prove that the category of fuzzy TL-uniform spaces and uniform maps between them is a topological category, by constructing optimal lifts of sources in it. For expansion on the categorical notions mentioned below, see [15]. The category FTS of fuzzy topological spaces and their continuous functions is a topological category. This means that its objects are sets with structures (here, fuzzy topologies), its morphisms are admissible functions between those sets (here, the continuous functions), morphism composition and the identity morphisms are the usual ones for functions, and the following three conditions are satisfied for all nonempty sets X. (1) Every source ( f j : X → (X j ,σ j ) j ∈J ) of functions in the category FTS has a unique optimal lift (also called initial lift), namely, the coarsest structure on X making each function f j a morphism. Specifically, the optimal lift of that source is the fuzzy topology τ on X with subbasis [16]: τ ∗ = f j−1 (λ) ∈ I X : j ∈ J, λ ∈ σ j , (5.1) that is, τ is the smallest fuzzy topology on X that contains τ ∗ . (2) The class FTS(X), of all FTS-structures (i.e., fuzzy topologies) on X, is a set (smallness condition). (3) If X is a singleton, then FTS(X) is a singleton. An important construction is that of initial fuzzy TL-uniformities. Given a family ( f j : X → (X j ,μ j )) j ∈J , where for each j ∈ J, (X j ,μ j ) is a fuzzy TL-uniform space and f j is a function from some set X to X j , we want to construct on X (in accordance with the general definition of initial lifts in categories of sets with structures) a coarsest fuzzy TLuniformity making each function f j uniformly continuous. This fuzzy TL-uniformity is called the initial fuzzy TL-uniformity on X for the family ( f j : X → (X j ,μ j )) j ∈J . 22 Fuzzy TL-uniform spaces If (Y , ℘) is a fuzzy TL-uniform space and h : Y → X, then if X carries the initial fuzzy TL-uniformity, h is uniformly continuous if and only if each f j oh is uniformly continuous. Lemma 5.1. For a fuzzy TL-uniform base υ on a set Y and a function f : X → Y , ( f × f )−1 υ∼T ⊆ ( f × f )−1 (υ) ∼T ∼T = ( f × f )−1 (ϕ) : ϕ ∈ υ . (5.2) Proof. Let ψ ∈ ( f × f )−1 (υ∼T ) and γ ∈ I1 . Then there are ϕ ∈ υ∼T and ϕγ ∈ υ, with ψ = ( f × f )−1 (ϕ) and γTϕγ ≤ ϕ. By taking ψγ = ( f × f )−1 (ϕγ ) ∈ ( f × f )−1 (υ), we have γTψγ = γT( f × f )−1 ϕγ = ( f × f )−1 γTϕγ ≤ ( f × f )−1 (ϕ) = ψ. (5.3) Therefore, ψ ∈ [( f × f )−1 (υ)]∼T , which completes the proof of the lemma. Proposition 5.2. Let f : X → Y be a function. If υ is a fuzzy TL-uniform base on Y , then ( f × f )−1 (υ) is a fuzzy TL-uniform base on X. Proof. (FLUB1) ( f × f )−1 (υ) is indeed a prefilterbase. (FLUB2) For all ψ ∈ ( f × f )−1 (υ) and x ∈ X, we have ψ(x,x) = ( f × f )−1 (ϕ) (x,x) = ϕ f (x), f (x) for some ϕ ∈ υ (5.4) = 1. (FLUB3\ ) Let ψ ∈ ( f × f )−1 (υ) and γ ∈ I1 , we get ϕ and ϕγ in υ such that ψ = ( f × f )−1 (ϕ), γT ϕγ oT ϕγ ≤ s ϕ. (5.5) By taking ψγ = ( f × f )−1 (ϕγ ) ∈ ( f × f )−1 (υ), we have γT ψγ oT ψγ = γT ( f × f )−1 ϕγ oT ( f × f )−1 ϕγ = γT ( f × f )−1 ϕγ oT ϕγ ≤ ( f × f ) −1 s ϕ (5.6) = s ψ, which completes the proof that ( f × f )−1 (υ) is a fuzzy TL-uniform base on X. By a similar proof, we also have the following proposition. K. A. Hashem and N. N. Morsi 23 Proposition 5.3. Let f j : X → Y j be a function, j ∈ J. If υ j is a fuzzy TL-uniform base on Y j for every j ∈ J, then j ∈J ( f × f )−1 (υ j ) is a fuzzy TL-uniform base on X, where j ∈J (f × f) −1 ⎧ ⎨0 υj = ⎩ (f × f) −1 ⎫ ⎬ ϕ j : J1 is a finite subset of J, ϕ j ∈ υ j ∀ j ⎭ . j ∈J1 (5.7) Theorem 5.4. A source in TL-FUS ( f j : X → (Y j ,μ j )) j ∈J has an optimal lift in TL-FUS, with fuzzy TL-uniform base υ= fj × fj −1 μj . j ∈J (5.8) Proof. Since ( f j × f j )−1 (μ j ) ⊆ j ∈J ( f j × f j )−1 (μ j ) = υ ⊆ υ∼T , then by Proposition 3.6, all f j are uniformly continuous: (X,υ∼T ) → (Y j ,μ j ). Let (Z, ℘) be in TL-FUS, and let a function h : Z → X be such that f j oh : (Z, ℘) → (Y j ,μ j ) are uniformly continuous for all j ∈ J. Then by Proposition 3.6 again, ℘⊇ = ℘⊇ f j oh × f j oh −1 f j × f j o(h × h) j ∈J (h × h)−1 ⎡ = (h × h)−1 ⎣ μj −1 fj × fj μj ∀ j ∈ J, that is, −1 fj × fj μj −1 j ∈J ⎤ (5.9) μj ⎦ = (h × h)−1 (υ). Referring to Proposition 3.6, once more, we find that h is uniformly continuous: (Z, ℘) → (X,υ∼T ). This establishes that (X,υ∼T ) is the optimal lift in TL-FUS of the given source. Proposition 5.5. TL-FUS is a topological category. Definition 5.6. The functor -t∼ : TL-FUS → T-FLS leaves the functions unaltered, and is defined on objects (X,μ) in TL-FUS by -t∼ (X,μ) = (X,τ(μ)). Theorem 5.7. The above -t∼ is a well-defined functor. The proof follows immediately from Theorem 3.10. Theorem 5.8. The functor -t∼ preserves optimal lifts. The proof can be analogously as [17, Theorem 7.3]. Theorem 5.9. Fuzzy TL-uniformizability is an initial property. Proof. This follows from Proposition 5.5 and Theorem 5.8. 24 Fuzzy TL-uniform spaces Acknowledgments The authors express their sincere thanks to the referees and the Area Editor for their thorough analysis of this manuscript and their most valuable suggestions which helped improve this manuscript greatly. References [1] E. P. Klement, R. Mesiar, and E. Pap, Triangular norms. Position paper. II. General constructions and parameterized families, Fuzzy Sets and Systems 145 (2004), no. 3, 411–438. [2] B. Schweizer and A. Sklar, Probabilistic metric spaces, Pacific Journal of Mathematics 10 (1960), 313–334. [3] N. N. Morsi, Hyperspace fuzzy binary relations, Fuzzy Sets and Systems 67 (1994), no. 2, 221– 237. , A small set of axioms for residuated logic, Information Sciences 175 (2005), no. 1-2, [4] 85–96. [5] L. A. Zadeh, Fuzzy sets as a basis for a theory of possibility, Fuzzy Sets and Systems 1 (1978), no. 1, 3–28. [6] S. Gottwald, Set theory for fuzzy sets of higher level, Fuzzy Sets and Systems 2 (1979), no. 2, 125– 151. [7] R. Lowen, Fuzzy topological spaces and fuzzy compactness, Journal of Mathematical Analysis and Applications 56 (1976), no. 3, 621–633. , A comparison of different compactness notions in fuzzy topological spaces, Journal of [8] Mathematical Analysis and Applications 64 (1978), no. 2, 446–454. , Fuzzy uniform spaces, Journal of Mathematical Analysis and Applications 82 (1981), [9] no. 2, 370–385. [10] N. N. Morsi, Fuzzy T-locality spaces, Fuzzy Sets and Systems 69 (1995), no. 2, 193–219. [11] B. Schweizer and A. Sklar, Probabilistic Metric Spaces, North-Holland Series in Probability and Applied Mathematics, North-Holland, New York, 1983. [12] U. Höhle, Probabilistic metrization of fuzzy uniformities, Fuzzy Sets and Systems 8 (1982), no. 1, 63–69. [13] A. Kandil, K. A. Hashem, and N. N. Morsi, A level-topologies criterion for Lowen fuzzy uniformizability, Fuzzy Sets and Systems 62 (1994), no. 2, 211–226. [14] U. Höhle, Probabilistic uniformization of fuzzy topologies, Fuzzy Sets and Systems 1 (1978), no. 4, 311–332. [15] J. Adámek, H. Herrlich, and G. E. Strecker, Abstract and Concrete Categories, Pure and Applied Mathematics (New York), John Wiley & Sons, New York, 1990. [16] R. Lowen, Initial and final fuzzy topologies and the fuzzy Tychonoff theorem, Journal of Mathematical Analysis and Applications 58 (1977), no. 1, 11–21. [17] K. A. Hashem and N. N. Morsi, Fuzzy T-neighbourhood spaces. II. T-neighbourhood systems, Fuzzy Sets and Systems 127 (2002), no. 3, 265–280. Khaled A. Hashem: Department of Mathematics, Faculty of Science, Benha University, Benha, Egypt E-mail address: khaledahashem@yahoo.com Nehad N. Morsi: Basic & Applied Science Department, Arab Academy for Sciences & Technology and Maritime Transport, P.O. Box 2033, Al-Horraya, Heliopolis, Cairo, Egypt E-mail address: nehad@aast.edu